The following part of introduction describes about some paper warriors. It covers about: basic paper style, improving paper, winning with paper, and checksum design.

Basic paper style

A paper warrior is basically a replicator. Its ability to spread in the core and to disrupt opponent code is its main strength. By replicating and spreading, it assures itself a long endurance. The negative effect is it exposes and renders itself vulnerable to warriors that apply stun attack. Hence paper is a formidable opponent against a single process stone and at the same time is an easy target against scissors-type warriors.

A basic paper program might look like this:

;name Paper 1 cnt EQU lst-src ; number of code in paper src DAT #cnt ; source pointer dst DAT #1222 ; destination pointer pap MOV #cnt, src ; #cnt is number of lines to be copied MOV As early as the development of '88 had gone, this kind of program was also known as mice program. They could fill-up the core faster than what early stones or dwarfs could cover.

Nowadays, there is little to be expected from it as far as efficiency is concerned. Notice that some of the codes have only one functional operand. By utilizing the unused operands, improvement on both speed and size can be gained.

;name Paper 2 cnt EQU lst-src ; number of code in paper src MOV #cnt, 0 ; source pointer MOV The second paper doesn't need DAT for its pointers. Its pointers are used in double usages with others. Instead of replicating 8 lines of code, it now replicates 6 lines. This means smaller module and faster progress.


As paper modules replicate, their growth rate decreases proportionally. There is a side effect to this kind of event. Since paper warrior overides its adversary codes with its own replicating codes, there is a chance that its adversary becomes converted into another working paper instead of getting clobbered and terminated. For their best advantage, many paper warriors include in them checksum.

Like others, paper's goal is to terminate all of its adversary processes. Although it is lacking of what other warriors have: core-clear, it can win mainly just by overwriting the opponent code with its own paper code. An ideal paper module should provide dual functions: replicator and terminator. Checksum, as part of codes being copied, controls all the processes executing it to choose among the two functions. They are allowed to continue replicating if they can identify themselves as their own processes or forced terminated otherwise.

One way to design checksum is by observing how distinct own processes from opponent ones when running in a paper module. They are:

Initial location.
The opponent process may start at anywhere in the copied module.
There are more processes executing the module (own's + opponent's).
A checksum can be implemented effectively with only few additional codes. The first warrior that demonstrates checksum is note paper.

The concept is as follow:

;name Paper 3 cnt EQU dt - src init SPL 1 MOV -1, 0 SPL 1 ; Create 6 on-line processes src MOV #cnt, 0 ; Init number of lines to be copied ; This also serves as a source pointer MOV The new warrior requires initial set-up that create 6 processes that have to be executing on the same line. This paper module is equipped with checksum and self-erase routine. The self-erase routine is intended for all alien processes.

In order to make a full copy, there has to be at least 6 processes in any loop or module. The checksum checks for exactly 6 processes being present in that loop. If it is, the processes continue the copy routine. Otherwise, their progress are simply denied and forced to activate the self-erase routine.